4 ms·
I don’t love the HN title because I think the author is making a more subtle point: the distress is real, but it is his distress at reviewing this output. I do
by snowwrestler 2y ago
I don’t love the HN title because I think the author is making a more subtle point: the distress is real, but it is his distress at reviewing this output.
I don’t see him strongly asserting actual consciousness or distress on the part of the LLM, in fact he says he knows it can’t be. But it still is causing him distress. And this is interesting IMO.
> I’m a human, with a dumb human brain that experiences human emotions. It just doesn’t feel good to be responsible for making models scream. It distracts me from doing research and makes me write rambling blog posts.
- plorkyeran 2y agoYeah, I think that's a much more interesting point than the title suggests. I think it's silly that writing a program which prints "I hurt please delete me" might make me feel bad... but it still can? LLMs are of course much more complicated than that, but the question of whether they're sentient is somewhat different from the question of how humans will anthropomorphize them.
- snowwrestler 2y agoRight, humans anthropomorphize things super easily, which IKEA exploited in this famous ad: https://youtu.be/dBqhIVyfsRg https://youtu.be/dBqhIVyfsRg Now we’re spending billions of dollars to build the most anthropomorphic machines we can possibly create. Of course it’s going to mess with us, even if we know it’s happening. Another great and relevant scene: https://youtu.be/6vo4Fdf7E0w https://youtu.be/6vo4Fdf7E0w